diff --git a/open-sse/config/providerModels.js b/open-sse/config/providerModels.js
index 68285d36ff..e0dbd7589d 100644
--- a/open-sse/config/providerModels.js
+++ b/open-sse/config/providerModels.js
@@ -213,6 +213,7 @@ export const PROVIDER_MODELS = {
   ],
   mmf: [  // MiMo Free — free channel only serves mimo-auto
     { id: "mimo-auto", name: "MiMo Auto" },
+    { id: "mimo-auto-claude", name: "MiMo Auto (Claude Compatible)", upstreamModelId: "mimo-auto" },
   ],
 
   cl: [  // Cline
diff --git a/open-sse/handlers/chatCore.js b/open-sse/handlers/chatCore.js
index 65281cfae2..1cbbef2475 100644
--- a/open-sse/handlers/chatCore.js
+++ b/open-sse/handlers/chatCore.js
@@ -15,7 +15,7 @@ import { getExecutor } from "../executors/index.js";
 import { buildRequestDetail, extractRequestConfig } from "./chatCore/requestDetail.js";
 import { handleForcedSSEToJson } from "./chatCore/sseToJsonHandler.js";
 import { handleNonStreamingResponse } from "./chatCore/nonStreamingHandler.js";
-import { handleStreamingResponse, buildOnStreamComplete } from "./chatCore/streamingHandler.js";
+import { handleStreamingResponse, buildOnStreamComplete, handleClaudeJsonAsStreamingResponse, handleOpenAIJsonAsClaudeStreamingResponse } from "./chatCore/streamingHandler.js";
 import { detectClientTool, isNativePassthrough } from "../utils/clientDetector.js";
 import { dedupeTools } from "../utils/toolDeduper.js";
 import { injectCaveman } from "../rtk/caveman.js";
@@ -60,7 +60,16 @@ export async function handleChatCore({ body, modelInfo, credentials, log, onCred
 
   const clientRequestedStreaming = body.stream === true || sourceFormat === FORMATS.ANTIGRAVITY || sourceFormat === FORMATS.GEMINI || sourceFormat === FORMATS.GEMINI_CLI;
   const providerRequiresStreaming = provider === "openai" || provider === "codex" || provider === "commandcode";
-  let stream = providerRequiresStreaming ? true : (body.stream !== false);
+  const mmfClaudeSyntheticNeeded = provider === "mmf" && sourceFormat === FORMATS.CLAUDE && body.stream === true && (
+    (Array.isArray(body.tools) && body.tools.length >= 20) ||
+    (Array.isArray(body.messages) && body.messages.length >= 12) ||
+    JSON.stringify(body).length >= 200000
+  );
+  const synthesizeClaudeStream = sourceFormat === FORMATS.CLAUDE && body.stream === true && (
+    (provider === "xiaomi-tokenplan" && modelTargetFormat === FORMATS.CLAUDE) ||
+    mmfClaudeSyntheticNeeded
+  );
+  let stream = synthesizeClaudeStream ? false : (providerRequiresStreaming ? true : (body.stream !== false));
 
   // DeepSeek-TUI: interactive TUI panel sends stream:true and needs SSE.
   // Non-interactive mode (-p flag) sends without stream and can't parse SSE.
@@ -268,6 +277,16 @@ export async function handleChatCore({ body, modelInfo, credentials, log, onCred
     if (result) { streamController.handleComplete(); return result; }
   }
 
+  // Streaming response synthesized from a non-streaming upstream response.
+  // This keeps Claude Code on SSE while avoiding flaky provider stream sockets.
+  const { onStreamComplete } = buildOnStreamComplete({ ...sharedCtx });
+  if (synthesizeClaudeStream) {
+    if (targetFormat === FORMATS.OPENAI) {
+      return handleOpenAIJsonAsClaudeStreamingResponse({ ...sharedCtx, providerResponse, reqLogger, streamController, onStreamComplete, trackDone, appendLog });
+    }
+    return handleClaudeJsonAsStreamingResponse({ ...sharedCtx, providerResponse, reqLogger, streamController, onStreamComplete, trackDone, appendLog });
+  }
+
   // True non-streaming response
   if (!stream) {
     const result = await handleNonStreamingResponse({ ...sharedCtx, providerResponse, sourceFormat, targetFormat, reqLogger, toolNameMap, trackDone, appendLog });
@@ -276,7 +295,6 @@ export async function handleChatCore({ body, modelInfo, credentials, log, onCred
   }
 
   // Streaming response
-  const { onStreamComplete } = buildOnStreamComplete({ ...sharedCtx });
   return handleStreamingResponse({ ...sharedCtx, providerResponse, sourceFormat, targetFormat, userAgent, reqLogger, toolNameMap, streamController, onStreamComplete });
 }
 
diff --git a/open-sse/handlers/chatCore/streamingHandler.js b/open-sse/handlers/chatCore/streamingHandler.js
index f96bb9bd06..e154a06fe1 100644
--- a/open-sse/handlers/chatCore/streamingHandler.js
+++ b/open-sse/handlers/chatCore/streamingHandler.js
@@ -8,6 +8,171 @@ import { buildAbortedResponsesTerminalBytes } from "../../utils/responsesStreamH
 import { buildRequestDetail, extractRequestConfig, saveUsageStats } from "./requestDetail.js";
 import { saveRequestDetail } from "@/lib/usageDb.js";
 
+const encoder = new TextEncoder();
+
+function sseEvent(event, data) {
+  return `event: ${event}\ndata: ${JSON.stringify(data)}\n\n`;
+}
+
+function normalizeClaudeUsage(usage = {}) {
+  const inputTokens = usage.input_tokens || usage.prompt_tokens || 0;
+  const outputTokens = usage.output_tokens || usage.completion_tokens || 0;
+  const cacheRead = usage.cache_read_input_tokens || 0;
+  const cacheCreate = usage.cache_creation_input_tokens || 0;
+  const normalized = {
+    prompt_tokens: inputTokens,
+    completion_tokens: outputTokens,
+    total_tokens: inputTokens + outputTokens,
+    input_tokens: inputTokens,
+    output_tokens: outputTokens
+  };
+  if (cacheRead > 0) normalized.cache_read_input_tokens = cacheRead;
+  if (cacheCreate > 0) normalized.cache_creation_input_tokens = cacheCreate;
+  return normalized;
+}
+
+function openAIJsonToClaudeJson(responseBody, model) {
+  const choice = responseBody?.choices?.[0] || {};
+  const message = choice.message || {};
+  const content = [];
+
+  if (message.content) {
+    content.push({ type: "text", text: String(message.content) });
+  }
+
+  if (Array.isArray(message.tool_calls)) {
+    for (const toolCall of message.tool_calls) {
+      const fn = toolCall?.function || {};
+      let input = {};
+      if (typeof fn.arguments === "string" && fn.arguments.trim()) {
+        try { input = JSON.parse(fn.arguments); } catch { input = { arguments: fn.arguments }; }
+      } else if (fn.arguments && typeof fn.arguments === "object") {
+        input = fn.arguments;
+      }
+      content.push({
+        type: "tool_use",
+        id: toolCall.id || `toolu_${Date.now()}_${content.length}`,
+        name: fn.name || "tool",
+        input
+      });
+    }
+  }
+
+  const usage = responseBody?.usage || {};
+  const promptTokens = usage.prompt_tokens || 0;
+  const completionTokens = usage.completion_tokens || 0;
+  const cacheRead = usage.cache_read_input_tokens || usage.prompt_tokens_details?.cached_tokens || 0;
+  const cacheCreate = usage.cache_creation_input_tokens || usage.prompt_tokens_details?.cache_creation_tokens || 0;
+  const claudeUsage = {
+    input_tokens: Math.max(0, promptTokens - cacheRead - cacheCreate),
+    output_tokens: completionTokens
+  };
+  if (cacheRead > 0) claudeUsage.cache_read_input_tokens = cacheRead;
+  if (cacheCreate > 0) claudeUsage.cache_creation_input_tokens = cacheCreate;
+
+  return {
+    id: responseBody?.id || `msg_${Date.now()}`,
+    type: "message",
+    role: "assistant",
+    model: responseBody?.model || model,
+    content,
+    stop_reason: choice.finish_reason === "tool_calls" ? "tool_use" : choice.finish_reason === "length" ? "max_tokens" : "end_turn",
+    stop_sequence: null,
+    usage: claudeUsage
+  };
+}
+
+function claudeJsonToSSEStream(responseBody, model) {
+  return new ReadableStream({
+    start(controller) {
+      const id = responseBody?.id || `msg_${Date.now()}`;
+      const responseModel = responseBody?.model || model;
+      const usage = responseBody?.usage || {};
+      const content = Array.isArray(responseBody?.content) ? responseBody.content : [];
+
+      controller.enqueue(encoder.encode(sseEvent("message_start", {
+        type: "message_start",
+        message: {
+          id,
+          type: "message",
+          role: "assistant",
+          model: responseModel,
+          content: [],
+          stop_reason: null,
+          stop_sequence: null,
+          usage: {
+            input_tokens: usage.input_tokens || 0,
+            output_tokens: 0,
+            ...(usage.cache_read_input_tokens ? { cache_read_input_tokens: usage.cache_read_input_tokens } : {}),
+            ...(usage.cache_creation_input_tokens ? { cache_creation_input_tokens: usage.cache_creation_input_tokens } : {})
+          }
+        }
+      })));
+
+      content.forEach((block, index) => {
+        if (block?.type === "text") {
+          controller.enqueue(encoder.encode(sseEvent("content_block_start", {
+            type: "content_block_start",
+            index,
+            content_block: { type: "text", text: "" }
+          })));
+          if (block.text) {
+            controller.enqueue(encoder.encode(sseEvent("content_block_delta", {
+              type: "content_block_delta",
+              index,
+              delta: { type: "text_delta", text: block.text }
+            })));
+          }
+          controller.enqueue(encoder.encode(sseEvent("content_block_stop", { type: "content_block_stop", index })));
+        } else if (block?.type === "thinking") {
+          controller.enqueue(encoder.encode(sseEvent("content_block_start", {
+            type: "content_block_start",
+            index,
+            content_block: { type: "thinking", thinking: "", signature: block.signature || "" }
+          })));
+          if (block.thinking) {
+            controller.enqueue(encoder.encode(sseEvent("content_block_delta", {
+              type: "content_block_delta",
+              index,
+              delta: { type: "thinking_delta", thinking: block.thinking }
+            })));
+          }
+          controller.enqueue(encoder.encode(sseEvent("content_block_stop", { type: "content_block_stop", index })));
+        } else if (block?.type === "tool_use") {
+          controller.enqueue(encoder.encode(sseEvent("content_block_start", {
+            type: "content_block_start",
+            index,
+            content_block: { type: "tool_use", id: block.id, name: block.name, input: {} }
+          })));
+          const inputJson = JSON.stringify(block.input || {});
+          if (inputJson && inputJson !== "{}") {
+            controller.enqueue(encoder.encode(sseEvent("content_block_delta", {
+              type: "content_block_delta",
+              index,
+              delta: { type: "input_json_delta", partial_json: inputJson }
+            })));
+          }
+          controller.enqueue(encoder.encode(sseEvent("content_block_stop", { type: "content_block_stop", index })));
+        }
+      });
+
+      const stopReason = responseBody?.stop_reason || "end_turn";
+      controller.enqueue(encoder.encode(sseEvent("message_delta", {
+        type: "message_delta",
+        delta: { stop_reason: stopReason, stop_sequence: responseBody?.stop_sequence || null },
+        usage: {
+          output_tokens: usage.output_tokens || 0,
+          ...(usage.input_tokens ? { input_tokens: usage.input_tokens } : {}),
+          ...(usage.cache_read_input_tokens ? { cache_read_input_tokens: usage.cache_read_input_tokens } : {}),
+          ...(usage.cache_creation_input_tokens ? { cache_creation_input_tokens: usage.cache_creation_input_tokens } : {})
+        }
+      })));
+      controller.enqueue(encoder.encode(sseEvent("message_stop", { type: "message_stop" })));
+      controller.close();
+    }
+  });
+}
+
 const SSE_HEADERS = {
   "Content-Type": "text/event-stream",
   "Cache-Control": "no-cache",
@@ -105,3 +270,112 @@ export function buildOnStreamComplete({ provider, model, connectionId, apiKey, r
 
   return { onStreamComplete, streamDetailId };
 }
+/**
+ * Handle a non-streaming Claude JSON provider response while preserving a Claude SSE client contract.
+ * Used for providers whose upstream streaming path is unstable but whose non-streaming endpoint is reliable.
+ */
+export async function handleClaudeJsonAsStreamingResponse({ providerResponse, provider, model, body, stream, translatedBody, finalBody, requestStartTime, connectionId, apiKey, clientRawRequest, onRequestSuccess, reqLogger, streamController, onStreamComplete, trackDone, appendLog }) {
+  let responseBody;
+  try {
+    responseBody = await providerResponse.json();
+  } catch (err) {
+    streamController.handleError(err);
+    appendLog?.({ status: "FAILED 502" });
+    return {
+      success: false,
+      response: new Response(JSON.stringify({ error: { message: `Invalid JSON response from ${provider}` } }), {
+        status: 502,
+        headers: { "Content-Type": "application/json", "Access-Control-Allow-Origin": "*" }
+      })
+    };
+  }
+
+  reqLogger?.logProviderResponse?.(providerResponse.status, providerResponse.statusText, providerResponse.headers, responseBody);
+  if (onRequestSuccess) await onRequestSuccess();
+
+  const usage = normalizeClaudeUsage(responseBody?.usage || {});
+  appendLog?.({ tokens: usage, status: "200 OK" });
+
+  const textContent = Array.isArray(responseBody?.content)
+    ? responseBody.content.filter(b => b?.type === "text").map(b => b.text || "").join("")
+    : "";
+  const thinking = Array.isArray(responseBody?.content)
+    ? responseBody.content.filter(b => b?.type === "thinking").map(b => b.thinking || "").join("")
+    : null;
+  onStreamComplete?.({ content: textContent, thinking }, usage, Date.now());
+  trackDone?.();
+  streamController.handleComplete();
+
+  saveRequestDetail(buildRequestDetail({
+    provider, model, connectionId,
+    latency: { ttft: Date.now() - requestStartTime, total: Date.now() - requestStartTime },
+    tokens: usage,
+    request: extractRequestConfig(body, stream),
+    providerRequest: finalBody || translatedBody || null,
+    providerResponse: responseBody || null,
+    response: { content: textContent, thinking, type: "streaming" },
+    status: "success"
+  }, { endpoint: clientRawRequest?.endpoint || null })).catch(err => {
+    console.error("[RequestDetail] Failed to save synthetic stream:", err.message);
+  });
+
+  return {
+    success: true,
+    response: new Response(claudeJsonToSSEStream(responseBody, model), { headers: SSE_HEADERS })
+  };
+}
+
+
+/**
+ * Handle a non-streaming OpenAI JSON provider response while preserving a Claude SSE client contract.
+ * Used for MMF free Claude mode because its upstream SSE socket can stall after first bytes.
+ */
+export async function handleOpenAIJsonAsClaudeStreamingResponse({ providerResponse, provider, model, body, stream, translatedBody, finalBody, requestStartTime, connectionId, apiKey, clientRawRequest, onRequestSuccess, reqLogger, streamController, onStreamComplete, trackDone, appendLog }) {
+  let responseBody;
+  try {
+    responseBody = await providerResponse.json();
+  } catch (err) {
+    streamController.handleError(err);
+    appendLog?.({ status: "FAILED 502" });
+    return {
+      success: false,
+      response: new Response(JSON.stringify({ error: { message: `Invalid JSON response from ${provider}` } }), {
+        status: 502,
+        headers: { "Content-Type": "application/json", "Access-Control-Allow-Origin": "*" }
+      })
+    };
+  }
+
+  reqLogger?.logProviderResponse?.(providerResponse.status, providerResponse.statusText, providerResponse.headers, responseBody);
+  if (onRequestSuccess) await onRequestSuccess();
+
+  const claudeBody = openAIJsonToClaudeJson(responseBody, model);
+  const usage = normalizeClaudeUsage(claudeBody.usage || {});
+  appendLog?.({ tokens: usage, status: "200 OK" });
+
+  const textContent = Array.isArray(claudeBody.content)
+    ? claudeBody.content.filter(b => b?.type === "text").map(b => b.text || "").join("")
+    : "";
+  const thinking = null;
+  onStreamComplete?.({ content: textContent, thinking }, usage, Date.now());
+  trackDone?.();
+  streamController.handleComplete();
+
+  saveRequestDetail(buildRequestDetail({
+    provider, model, connectionId,
+    latency: { ttft: Date.now() - requestStartTime, total: Date.now() - requestStartTime },
+    tokens: usage,
+    request: extractRequestConfig(body, stream),
+    providerRequest: finalBody || translatedBody || null,
+    providerResponse: responseBody || null,
+    response: { content: textContent, thinking, type: "streaming" },
+    status: "success"
+  }, { endpoint: clientRawRequest?.endpoint || null })).catch(err => {
+    console.error("[RequestDetail] Failed to save synthetic OpenAI stream:", err.message);
+  });
+
+  return {
+    success: true,
+    response: new Response(claudeJsonToSSEStream(claudeBody, model), { headers: SSE_HEADERS })
+  };
+}
diff --git a/open-sse/translator/response/openai-to-claude.js b/open-sse/translator/response/openai-to-claude.js
index 6a46551d65..34779ba7d6 100644
--- a/open-sse/translator/response/openai-to-claude.js
+++ b/open-sse/translator/response/openai-to-claude.js
@@ -40,6 +40,11 @@ function isValidPdfPagesArg(filePath, pages) {
     /^\d+(?:-\d+)?$/.test(pages);
 }
 
+function shouldSuppressReasoningContent(state) {
+  const provider = String(state?.provider || "").toLowerCase();
+  return provider.startsWith("opencode") || provider === "mimo-free" || provider === "mmf";
+}
+
 // Helper: stop thinking block if started
 function stopThinkingBlock(state, results) {
   if (!state.thinkingBlockStarted) return;
@@ -131,7 +136,7 @@ export function openaiToClaudeResponse(chunk, state) {
 
   // Handle reasoning_content (thinking) - GLM, DeepSeek, etc.
   const reasoningContent = delta?.reasoning_content || delta?.reasoning;
-  if (reasoningContent) {
+  if (reasoningContent && !shouldSuppressReasoningContent(state)) {
     stopTextBlock(state, results);
 
     if (!state.thinkingBlockStarted) {